Auditory profile analysis of irregular sound spectra

Area of Science:

• Auditory perception
• Psychoacoustics
• Signal processing

Background:

• Sound spectral shape discrimination is crucial for auditory perception.
• Listeners' ability to detect subtle changes in complex sounds is key to understanding auditory processing.
• Previous research has explored auditory profile analysis, but the impact of spectral variability requires further investigation.

Purpose of the Study:

• To investigate the learning process involved in discriminating changes in sound spectral shapes.
• To examine how random variations in spectral patterns affect the ability to detect intensity changes.
• To provide insights into the roles of critical band filtering, sensory variability, and memory in auditory perception.

Main Methods:

• Naive listeners were tested on their ability to discriminate intensity increments in a specific frequency component (948 Hz) of a multitone complex.
• Learning was assessed over 3000 trials with a regular reference spectrum.
• Subsequent experiments involved randomly perturbed spectral shapes, varying frequencies and amplitudes, with listeners practicing on specific random spectra.

Main Results:

• Listener thresholds improved by approximately 10 dB over 3000 trials, with substantial learning occurring early in training.
• Randomizing spectral shapes, whether frequencies or amplitudes, presented challenges to discrimination.
• Performance varied depending on whether random spectra were fixed or varied across trials, highlighting the influence of spectral predictability.

Conclusions:

• Auditory perception of spectral changes involves learning and is influenced by spectral predictability and variability.
• Critical band filtering, sensory variability, and memory representations play significant roles in auditory profile analysis.
• The findings suggest limitations in generalizing simple energy detection models to complex auditory scene analysis.